Two competing systems, one Korean sound
Korean is written in Hangul, a script with no inherent relationship to the Latin alphabet, so any English rendering of a Korean name is necessarily a transliteration choice rather than a fixed fact. South Korea has used more than one official system for making that choice over the past century, and the switch between them is the direct source of a great deal of the spelling variation seen in Korean names today.
The older of the two systems in common use, McCune-Reischauer romanization, was developed by George M. McCune and Edwin O. Reischauer and first published in 1939. It aims to represent Korean pronunciation reasonably closely to an English-speaking reader, which requires diacritical marks — breves and apostrophes — to distinguish sounds that would otherwise collapse into the same plain Latin letters. South Korea adopted a modified version of McCune-Reischauer as its official government standard from 1984 until 2000, in part to standardize signage and documentation ahead of the 1988 Seoul Olympics. North Korea still uses a variant of McCune-Reischauer as its official system today.
South Korea replaced it on July 7, 2000, with Revised Romanization of Korean, issued by the Ministry of Culture and Tourism and now the country's official standard for government use, road signage, and place names. Revised Romanization was designed specifically to avoid diacritics and special characters, using only the plain 26-letter Latin alphabet so that names and place names would work cleanly with ordinary keyboards, databases, and international systems that could not reliably reproduce an apostrophe or a breve. That design goal is also exactly what produces the Kim/Gim disagreement.
Why the rule says Gim
Korean's consonant system does not map cleanly onto the English distinction between voiced and unvoiced stops (the difference an English speaker hears between, for instance, "g" and "k," or "b" and "p"). The Korean consonant at the start of 김 is what linguists describe as a lenis, unaspirated stop — a sound that does not correspond neatly to either the English "g" or the English "k," but that Revised Romanization's rules assign, by convention, to "g" when it opens a syllable. Applied mechanically and consistently, that rule renders 김 as "Gim," not "Kim." The same logic is why Revised Romanization spells the city Koreans and English speakers alike have long called "Pusan" as "Busan," and it is a large part of why the system drew criticism from linguists when it was introduced: applied strictly, it can produce spellings that read, to an English-speaking ear used to the older conventions, as if they represent the wrong sound entirely.
Revised Romanization is, in other words, internally consistent — it is applying one rule uniformly — but that rule was never going to match nearly a century of established, informal Latin-alphabet usage that had already settled on "Kim" long before the 2000 standard existed. "Kim" was not created by an older official romanization system so much as it was already the conventional English spelling in wide use well before any Korean government agency tried to standardize the process at all.
Why almost everyone writes Kim anyway
Personal and family names carry an explicit, acknowledged exception in Korean romanization practice that place names generally do not. Established surnames like Kim, Lee, and Park are so deeply entrenched in existing passports, English-language documents, business registrations, academic publications, and personal usage that the 2000 standard never displaced them in practice, despite being the official government rule. Surveys of the tiny minority who do use the technically correct "Gim" spelling put it at well under one percent of people who carry the surname — the remainder use "Kim," spellings close to it, or occasionally other historical variants such as "Ghim" or "Kin."
South Korean passport regulation reflects this directly rather than fighting it. The Enforcement Decree of the Passport Act does not require every surname to follow Revised Romanization; it permits romanization corrections and established alternative spellings under a defined set of categories, and Korean courts have weighed in on where the line sits. In one widely reported 2025 ruling, a Seoul court found that the government's romanization guidelines function as a non-binding guideline rather than an absolute legal requirement, and that authorities should generally accommodate an alternative spelling unless it creates a genuine risk to the passport's reliability. In a separate, later case, a different Seoul court reached the opposite result on a specific request to switch a passport surname from the conventional "Lee" to the technically-closer-to-pronunciation "Yi," ruling that a personal preference alone, without a documented practical hardship, was not sufficient grounds for the change. Read together, the two rulings describe a real, still-unsettled tension: Korean romanization has both an official rule and a strong, legally recognized tradition of established usage overriding it, and which one wins in a given case is not always predictable in advance.
One name, several spellings — and the reverse problem
The practical consequence for anyone trying to identify a specific person from a Korean name is two-directional. In one direction, a single Korean surname can be legitimately rendered several different ways in English — "Kim" overwhelmingly, but "Gim" and rarer variants exist on real documents and cannot simply be dismissed as errors. In the other direction, spellings that look identical in English do not always trace back to the same underlying Korean name or sound, because romanization is a many-to-one compression of a richer set of original Korean distinctions. Neither direction of this problem is unique to surnames.
The same pattern shows up constantly in Korean given names. The Korean given-name element commonly seen in English as "Jung," for instance, is romanized at least half a dozen different ways in real, current use — "Jung" itself is the most common choice on Korean passports, but "Jeong" (the Revised Romanization spelling), "Chung," "Jong," "Cheong," and several rarer variants all appear on real documents for what is, in Korean, the identical name. "Jong" specifically is the standard spelling under North Korea's official system rather than South Korea's, adding a geopolitical wrinkle to an already crowded field of options for a single sound.
Given-name substitution adds a third layer
Romanization variance is a spelling problem: the same underlying Korean name, written more than one way in the Latin alphabet. A separate and additional layer of variation comes from the practice, documented in a companion guide on this hub, of Korean immigrants and Korean-Americans adopting an English given name that is not a transliteration of their Korean name at all, but a separate name chosen alongside or in place of it — commonly for ease of pronunciation in an English-speaking environment, and, as that guide explains, disproportionately drawn from biblical names given Korea's Christian population. "John" functions this way for a range of different underlying Korean given names; it is not the English transliteration of any single one of them, the way "Gim" is a transliteration of 김. A person's legal or professional records might show "John" as a chosen English name with a completely different Korean given name behind it, a romanized version of a Korean given name that happens to resemble an English name, or, in rarer cases, an actual Korean rendering of the biblical John (요한, Yohan) used directly as a name by some Korean Christians. All three produce the identical English string "John" on a directory listing, a diploma, or a business card, with no indication from the spelling alone of which of the three is actually behind it.
Put together, romanization variance and given-name substitution mean that neither half of "John Kim" is a stable, one-to-one transliteration of a single underlying Korean identity. Both halves compress a wider range of original possibilities into one common English string — which is exactly why a name search or a directory listing built purely on English-language spelling can never, by itself, be trusted to distinguish one person from another who happens to share it.